An SDI-QRNG is a semi-device-independent quantum random number generator where the measurement device is considered trustworthy while the entropy source is not. This approach can eliminate eavesdropping entropy even when the entropy source of a general QRNG is potentially attacked, thereby ensuring the security and authenticity of random numbers generated using quantum resources. It has broad applications in fields such as cryptography, including but not limited to national security, financial transactions, cloud computing, and the Internet of Things. Typically, the working principle of SDI-QRNG can be described as: randomness is guaranteed by the intrinsic uncertainty of the quantum state, and security is achieved by estimating the random fluctuation deviation of the measurement basis selection, where randomness originates from the intrinsic uncertainty of quantum resources, and security is ensured by evaluating eavesdropping entropy. In this way, SDI-QRNG can generate secure random numbers without trusting the entropy source.

The main challenge in integrating SDI-QRNG on-chip is achieving both randomness and security. Firstly, for efficient detection and utilization of quantum randomness, it is crucial to achieve high-speed and distortion-free quantum resource on-chip detection, which determines how much quantum entropy the QRNG can obtain. More importantly, the security issue lies in the physical environment of the actual chip. Under the physical constraints at the chip level, defining the actual side information amount is extremely difficult and complex, which determines how much information in the random numbers remains secure. Due to the simultaneous existence of these two major challenges during monolithic integration, there has been no previous report on on-chip SDI-QRNG implementation.

To address this, the teams of Professor Guihua Zeng and Professor Kan Wu at Shanghai Jiao Tong University proposed and realized a novel on-chip source-device-independent quantum random number generation scheme. This scheme achieves high-fidelity randomness acquisition through high-bandwidth distortion-free coherent detection based on on-chip vacuum states and overcomes the practical security issues faced by on-chip SDI-QRNG due to inseparability by utilizing an improved entropy uncertainty principle mechanism under chip physical constraints. By maximizing the random fluctuation deviation of the measurement basis selection, the quantum secure entropy bound is accurately marked after eliminating eavesdropping entropy, ultimately ensuring both randomness and security of on-chip SDI-QRNG. This scheme achieves a secure bit generation rate of 146.2 Mbps within a photonic circuit size of 2.9 mm × 4.5 mm and a theoretical maximum coding rate of up to 248.47 Gbps. All extracted secure bits passed the NIST randomness test, demonstrating its simplicity and efficiency, and verifying the feasibility of on-chip SDI-QRNG implementation. This paves the way for further large-scale, low-cost, and secure deployment of quantum resource randomness at room temperature. Relevant research results were recently published in Photonics Research, Volume 12, Issue 7, 2024. [Lang Li, Minglu Cai, Tao Wang, Zicong Tan, Peng Huang, Kan Wu, Guihua Zeng, "On-chip source-device-independent quantum random number generator," Photonics Res. 12, 1379 (2024)]

Figure 1. Schematic of the integrated SDI-QRNG on SiPh PIC. GC, grating coupler; PM, phase modulator; Att, attenuator; MMI, multimode interference splitter; AWG, arbitrary waveform generator; PD, photodetector; DSO, digital signal oscilloscope. The input state is a vacuum state, and the laser is used for orthogonal component measurement.

Figure 1 shows the proposed on-chip SDI-QRNG scheme, which is based on coherent detection of vacuum states. To ensure the security of random number generation according to the Entropy Uncertainty Principle (EUP), the scheme achieves on-chip random measurement basis switching through PM modulation of LO. Ultimately, the SDI-QRNG chip can ensure both randomness and security.

Figure 2 presents the signal processing flow of the on-chip SDI-QRNG based on vacuum states. By randomly applying modulation voltage signals to verify the normal component generation, the non-reciprocal orthogonal component is acquired to define the quantum conditional minimum secure entropy after eliminating eavesdropping entropy.

Figure 2. Signal processing flow of the on-chip SDI-QRNG based on vacuum states. (a) Structure for generating secure random bits. (b) Modulation voltage signal for switching control of the measurement basis used for secure entropy estimation. When a switching control signal pulse is applied to select the measurement basis of the normal component Q̂ (gray dot), the verification data is obtained and used to estimate the bound of the actual on-chip quantum entropy $H_{min}^{\mathrm{pra}}(P_{\delta p}\mid E)$, which determines how much of the original random bits' remaining entropy is secure. By applying the strong randomness extractor $H_{min}^{\mathrm{pra}}(P_{\delta p}\mid E)$ calibrated according to the above protocol, the security of the original random bits is protected. Part of the secure random bits generated by the on-chip SDI-QRNG is fed back to enhance the randomness of the measurement basis selection.

Figure 3 shows the electron microscope image of the source-device-independent chip on SiPh PIC. The LO is coupled to the chip via a vertical GC from a 1550 nm narrow linewidth laser and then through a PM driven by AWG. The AWG provides secure random bits to control the PM when generating the selection signal of the verification orthogonal $Q_{\delta q}$ measurement basis. Meanwhile, the LO and vacuum state are split by a 50:50 MMI and enter the PD after interference. The PD output is monitored by the DSO and the differential current signal output is sampled by the oscilloscope.

Figure 3. Electron microscope image of the source-device-independent chip on SiPh PIC. (a) Overall microscopic view of the SDI-QRNG chip, including grating coupler (GC), phase modulator (PM), attenuator (Att), and photodetector (PD). The LO is a 1550 nm narrow linewidth laser, coupled to the chip via a vertical GC and then through a PM driven by AWG. The AWG provides secure random bits to control the PM when generating the selection signal of the verification orthogonal $Q_{\delta q}$ measurement basis. Meanwhile, the LO and vacuum state are split by a 50:50 MMI and enter the PD after interference. The PD output is monitored by the DSO and the differential current signal output is sampled by the oscilloscope. (b) Microscopic photo of the PD. The microscopic photo shows that the PD, based on Ge PIN PD, is 16 μm × 23 μm in size for light detection.

Figure 4 shows the experimental zero-difference measurement results of the vacuum state on the SiPh SDI-QRNG PIC, where Figure 4.(a) represents the quadrature components P and Q of the shot noise in each frame, and Figure 4.(b) represents the switching driving signal on the PM.

Figure 4. Experimental homodyne measurement of the vacuum state on the SiPh SDI-QRNG PIC. (a) Quadrature components P and Q of the shot noise in each frame. (b) Switching driving signal on the PM.

Figure 5 shows the quantum conditional minimum entropy of each sample under 250 kHz measurement basis switching. $H_{\text{low}}(P_{\delta }\mid E)$ represents the quantum conditional minimum entropy of each original sample under 250 kHz basis switching, and $H_{\text{low}}^{ϵ}(P_{{\delta }_{10}}\mid E)$ represents the quantum conditional minimum entropy of each original sample under 250 kHz considering the finite code length effect.

Figure 5. Quantum conditional minimum entropy of each sample under 250 kHz measurement basis switching. represents the quantum conditional minimum entropy of each original sample under 250 kHz basis switching, and represents the quantum conditional minimum entropy of each original sample under 250 kHz considering the finite code length effect.

Doctoral student Lang Li stated: "Semi-device-independent quantum security protocols are promising approaches that balance the security of device-independent protocols, the availability of general protocols, and the practical security of quantum resources. However, most previous schemes are complex and difficult to integrate, greatly limiting the large-scale deployment of the intrinsic randomness of quantum resources. The integrated source-device-independent secure scheme proposed in this paper, supported by a novel entropy uncertainty principle mechanism under chip physical constraints, can simultaneously overcome the challenges of high-bandwidth distortion-free detection of quantum randomness and precise estimation of the on-chip quantum conditional minimum secure entropy. The structure is compact and simple, verifying the feasibility of on-chip source-device-independent quantum random number generation. Due to its physical basis of vacuum state fluctuation interference, the chip theoretically achieves a bit rate of up to 248.47 Gbps, paving the way for further large-scale, low-cost, and secure deployment of quantum resource randomness at room temperature."